Post on 12-Jul-2015
Stephen Goldsmith
Harvard Kennedy School
Stephen Goldsmith
Harvard Kennedy School
SAP SAPPHIRE May 16, 2013
Use Data Better, Faster, and Cheaper to Drive Public Value
DRAFT -- not for distribution
2
1 Public sector organizations can’t solve complex horizontal problems with vertical solutions, nor by simply accomplishing bureaucratic activities better.
2 The role of the public sector is being transformed from direct service provider to generator of public value.
3 We won’t get the results taxpayers
deserve, nor citizens require, until
we figure out how to better
manage a public sector that uses
data better to hold itself and third
parties accountable.
Governing By Network: Leverage private sector, philanthropic, non-profit,
and citizen volunteers
DRAFT -- not for distribution
3
Problems Are No Longer Vertical
DRAFT -- not for distribution
Possible &
Practical
Everyone has mobile devices
Cloud computing reduces startup
costs
Proliferation of data and advanced tools
to integrate
Maturity in techniques to mine unstructured data
Budgetary pressures require efficiency
improvements
Analytics success in public & private sectors
The Perfect Technology Storm
4
DRAFT -- not for distribution
• Handheld devices can provide
decision support to field workers and
real-time supervision to managers
• Constituents as sensors: everyone
has mobile devices, allowing access
to information and GIS-stamped
reports and Tweets
Proliferation of Mobile Devices
5
DRAFT -- not for distribution
• Even smaller organizations can
access high-quality technology
through SaaS
• Organizations can get out of the
infrastructure business and into the
solution business
Cloud Computing
6
DRAFT -- not for distribution Data Mining: HR as an Example
7
• Electronic time and attendance systems produce
more than compliance information.
• Field workers—teachers, bus drivers, street and
public works employees, public safety, child welfare—
are the backbone of our society.
• Connecting digital time and attendance to work flow
systems allows insights into information.
Overtime scheduling
Time on task
Productivity analysis
Route and labor assignments
Time on task by division, area and worker
DRAFT -- not for distribution Unstructured Data: Social Media Mining
Grade.DC.Gov
• Distills various forms of customer feedback into a letter grade for each agency
• Uses an algorithm that takes into account comments submitted through the website, texts, and via social media
• Social media mining captures input from a broader set of customers.
8
DRAFT -- not for distribution Data Analytics
• Predictive Analyses
By highlighting common issues before they occur
Question: What factors make a building most at risk for fires?
• Root Cause Analyses
By providing insights that explain common incidents
Question: Why are there frequent accidents at certain intersections? Which
individuals best benefit from job training?
• Increased Accountability
By monitoring areas for improvement
Question: Which inspectors are behind schedule?
• Improved Operational Management
By providing data-driven solutions to promote more effective business processes
Question: What are the best routes for vehicles to take?
9
DRAFT -- not for distribution Using Data Analytics:
Spend Management / Do Not Pay
DRAFT -- not for distribution Using Data Analytics:
Revenue Discovery
11
DRAFT -- not for distribution Using Data Analytics:
Fraud Waste and Abuse
12
DRAFT -- not for distribution
The University of Kentucky uses a real-time data platform from SAP to integrate data sources, improve student retention through earlier intervention, and gain insight into finances and resource use.
Using Data Analytics: Higher Education
DRAFT -- not for distribution Performance, Accountability,
Transparency
14
Data Mining - HANA • Automated Voice Response • Internet/Web • Mobile phone/data/picture • Bar Code/GPS • Point of Service Device
Alerts/Dashboards/Decision
Support for Outcomes • Caseworker • Court/law enforcement • Supervisor • Caregiver
Billing/Epayment/EBT
Device Agnostic
Guardian Counselor Previous
Cases School Police Medical 15
Child Welfare Example
Decision Support
Caseworker
DRAFT -- not for distribution
Reduce risks and improve outcomes for children in
need of services
• Monitor required daily events
• Provide immediate alerts for potentially dangerous
situations
• Increase the integrity of visits
• Increase the quality of discretionary decisions through
the application of decision support
Improve caseworker tools, processes, and efficiency
• Simplify and automate communication and
documentation in the field
• Provide caseworkers “exception reports” when ordered
events did not occur
• Engage network of community stakeholders
• Reduce inefficiency and red tape by eliminating millions
of pieces of paper
Goals
1
2
16
DRAFT -- not for distribution
Frequent, reportable events • Account for every child every day (via school, daycare, etc.)
• Verify monthly caseworker visits
• Minimize time child is unaccounted for
• Enlist help of stakeholders throughout the community
• Report monthly tracking results to caseworkers
Proactive alerting and intervention • Respond to irregularities and missed events before crises occur
• Automate escalation to management and stakeholders
• Disseminate immediate risk alerts to appropriate responders
Visit verification • Track date, time, and GPS location for every contact
• Photograph child during each visit
• Identify in-home individual
Enhanced data collection and use • Maximize federal funding
• Replace millions of pieces of paper with secure electronic data
Core Child Safety Concept
17
DRAFT -- not for distribution
• The routine is automated—the smart field worker
• Discretion and accountability both possible
• Work vertically; think horizontally
• In Rem memory increases problem solving
• Predictive analytics allow problem preemption
• Performance measurement, transparency and stat programs at a new level
• The 20% rule—everything can be improved 20%:
Children’s well-being
Tax revenue recognized
Risks cured, accidents prevented
Waste, fraud and abuse reduced
The Newly Empowered Worker
18
DRAFT -- not for distribution
Learn more at www.datasmartcitysolutions.org